Systems and methods ensure that datagrams retain integrity in light of the problems associated with the internetworking protocol's use of datagram identifiers drawn from a first pool of datagram identifiers. The methods involve controlling the use of datagram identifiers to ensure that only certain identifiers are allowable. A second pool of datagram identifiers is created that is different from the first pool. A datagram identifier is drawn from the second pool and assigned to a datagram in a manner that the datagram identifier is controlled from being reused during the lifetime of the datagram. In alternative embodiments the number of available datagram identifiers is either increased through optional header fields, or reduced through tracking allowed datagram identifiers. A first host notifies a second host of an allowed identifier. The second host uses the allowed identifier as a datagram identifier.
|
1. A method comprising:
selecting, by a first host device, a datagram identifier from a list of available datagram identifiers;
placing, by the first host device, the datagram identifier in a list of unavailable datagram identifiers;
transmitting, by the first host device, a datagram request to a second host device over a communication link, the datagram request including the datagram identifier;
receiving, by the first host device, a datagram response from the second host device including the datagram identifier to identify a datagram, the datagram response further including a fragment of the datagram;
removing, by the first host device, the datagram identifier from the list of unavailable datagram identifiers based at least in part on receipt of the datagram; and
placing, by the first host device, the datagram identifier in the list of available datagram identifiers based at least in part on said removing.
6. A computing device comprising:
a controller configured
to select a datagram identifier from a list of available datagram identifiers;
to place the datagram identifier in a list of unavailable datagram identifiers;
to generate a datagram request including the datagram identifier;
to remove the datagram identifier from the list of unavailable datagram identifiers based at least in part on receipt of a datagram to which the datagram identifier is associated; and
to place the datagram identifier in the list of available datagram identifiers when the datagram identifier is removed from the list of unavailable datagram identifiers; and
a transceiver coupled to the controller and configured
to transmit the datagram request to another computing device via a communication link; and
to receive a datagram response, which includes the datagram identifier and a fragment of the datagram, from the another computing device via the communication link.
2. The method of
generating, by the first host device, the datagram request in a manner such that the datagram identifier is included in a payload of the datagram request.
3. The method of
4. The method of
receiving a plurality of datagram responses including a corresponding plurality of fragments of the datagram;
assembling the plurality of fragments into the datagram; and
removing the datagram identifier from the list of unavailable datagram identifiers and placing the datagram identifier in the list of available datagram identifiers based at least in part on said assembling.
5. The method of
determining a timeout event associated with another datagram identifier in the list of unavailable datagram identifiers; and
removing the another datagram identifier from the list of unavailable datagram identifiers and placing the another datagram identifier in the list of available datagram identifiers based at least in part on said determining of the timeout event.
7. The computing device of
8. The computing device of
9. The computing device of
assemble a plurality of fragments, including the fragment, into the datagram; and
remove the datagram identifier from the list of unavailable datagram identifiers based at least further in part on a successful assembly of the datagram.
|
|||||||||||||||||||||||||||
This patent application claims priority to provisional patent application Ser. No. 60/791,051 filed on Apr. 10, 2006.
The field of invention relates to reliably transferring a datagram over an internetworking protocol.
On Apr. 4, 2006, a notification was published referencing a problem with the standard internetworking protocol Ipv4 as defined in RFC 791 maintained by the Internet Engineering Task Force (IETF). The reference written by J. Heffner, M. Mathis, and B. Chandler (herein incorporated by reference in its entirety) describes how datagrams transferred over Ipv4 can suffer data corruption due to issues relating to the datagram identification field within an Ipv4 header. The datagram corruption results from having a limited number of available datagram identifiers during the lifetime of the datagram. No solution to the problem was offered by the authors.
IPv4 can transfer a datagram having a size up to 65,535 (216) bytes in size. An Internetworking Protocol (IP) layer of a protocol stack including IPv4 can assign the datagram a 16-bit identifier which implies there are 65,535 possible datagram identifiers. Typically, an IP datagram identifier is implemented as a counter that is incremented every time a datagram identifier is used. When the counter reaches its maximum value, 65,535 for a 16-bit counter, the value returns to zero. The IP will fragment a large datagram into smaller chunks to send the fragments over a media, Ethernet for example. If Ethernet supports 1500 byte frames, then the IP layer will create up to 44 frames where each frame has the same datagram identifier and has an offset into the datagram. The datagram identifier and offset information are used by a remote host to reassemble the datagram before passing the datagram to the upper layers of the communication stack. For high speed media a host could send more than 65,535 datagrams in a very short time causing the host to wrap the value of the datagram identifier counter. For example, if Ethernet running at 1 Gigabit per second (Gbps) is used, the datagram identifier counter could wrap in less than 1 second assuming relatively small datagrams. Most communication stacks hold a datagram for reassembly from 30 seconds to 120 seconds. Therefore, if a datagram having a specific datagram identifier is stored in memory for reassembly and one of its fragments is lost, then a subsequent, different datagram fragment having the same identifier could cause a corruption of the first datagram. The corruption occurs because the receiving host interprets the fragment from the second datagram as belonging to the first datagram because it has the same datagram identifier and offset. This problem applies to TCP, UDP, ICMP, or other data transported over IP. The problem can be characterized as resulting from using a pool of datagram identifiers where the pool has a limited number of available datagram identifiers while the datagram is alive in the system.
Interestingly, Zetera™ Corporation, a producer of network storage technology, encountered and resolved the datagram corruption problem in the same time frame as the public release of the problem statement. Zetera discovered the problem while running a proprietary storage protocol, the Z-SAN™ protocol, over UDP/IP on a 1 Gbps Ethernet system. Zetera has created a solution that solves the problem as described in Zetera U.S. provisional patent application assigned Ser. No. 60/791,051 field on Apr. 10, 2006 herein incorporated by reference in its entirety.
Further research regarding the datagram corruption issue indicates that the problem has manifested itself as far back as 1987 when customers of Sun's Network File System (NFS) implementation suffered from data corruption. NFS used 8 KB UDP datagrams that would become corrupted for the reason described above. Implementations of NFS addressed the problem by shortening the time NFS waits for a response at the application layer, or through large checksum values on the datagram (32-bit checksums or greater). Such solutions mitigate the risk of loss, but do not create a solution for the problem. In addition, short timeouts reduce efficiency of the system because the system must conduct additional retries.
U.S. Pat. No. 6,894,976 titled “Prevention and detection of IP identification wraparound errors” teaches a method of reducing the risk associated with the problem of IP datagram identifier wrap around through the use of timeouts and checksums. However, this reference does not present a viable solution for the problem that applies to all IP based applications.
The described problem is an inherent part of standardized IPs and can not be resolved universally without changing the standard. However, it is desirable to have a real solution that can resolve the issue in a manner that applies to network storage, other network devices, or network applications. It is contemplated that a real solution would be adopted by the standards. A desirable solution should have the following characteristics:
Clearly there remains a long felt need for a solution to the datagram corruption problem. Preferably the solution completely addresses the problem rather than simply mitigating the risk of the problem occurring.
The inventive subject matter comprises methods for resolving a datagram corruption problem that results from using datagram identifiers from a pool of identifiers that have a limited supply. The methods include providing a pool of identifiers that is different than the first pool and assigning a datagram identifier drawn from the second pool to a datagram. Furthermore, the datagram identifier is controlled from being reused during the lifetime of the datagram to ensure the corruption problem does not arise. In some embodiments, the second pool of datagram identifiers comprises a larger number of datagram identifiers and in other embodiments the second pool comprises a smaller number of datagram identifiers. In preferred embodiments where the number of available datagram identifiers is increased, the time in which the increased number of identifiers is consumed exceeds the time period in which the datagram corruption problem is likely to occur in light of the datagram transfer rate. In other preferred embodiments where the number of available datagram identifiers is reduced, especially down to a single identifier, the identifier is controlled from being reused with subsequent datagrams.
Especially preferred embodiments comprise storage protocols that utilize the methods disclosed. Data is exchanged with a storage device using the storage protocols via the contemplated datagrams. Storage protocols include proprietary storage protocols or standard storage protocols. Examples of standard storage protocols include iSCSI, FCIP, or iFCP.
In some embodiments, a first host keeps track of which datagram identifiers are currently in use. The first host notifies a second host of a datagram identifier that is allowable; then the second host uses the datagram identifier for a datagram sent back to the first host. The second host will not use the datagram identifier again unless the first host indicates that it can be reused. The first host tracks used datagram identifiers through a list. The datagram identifier is associated with the datagram in manner that communication protocol stacks are able to reassemble fragmented datagrams as they normally would. Datagram identifiers that have been recently used are controlled from being used again to ensure that following datagrams do not corrupt the reassembly of the current datagram.
The teachings herein may be advantageously employed by developers of networking applications that use internetworking protocols to resolve datagram corruption that can cause data corruption. The methods presented can be used in conjunction with TCP or UDP transports or other protocols that leverage IP for sending datagrams. The methods are especially useful for deployment in network devices accessible by clients including data storage applications where data integrity is a paramount concern.
Glossary
The following descriptions refer to terms used within this document. The terms are provided to ensure clarity when discussing the various aspects of the invention matter, without implied limitation, and to reduce repetition within the body of the document.
The term “datagram” means a chunk of data sent over a communications link where the chunk can be fragmented into smaller pieces for ease of delivery. Typically, datagrams are associated with internetworking protocols (IP) including IPv4 or IPv6. In addition a datagram is identified through a datagram identifier to aid in reassembly of the fragments.
Various objects, features, aspects, and advantages of the present invention will become more apparent from the following detailed description of the preferred embodiments of the invention, along with the accompanying drawings in which like numerals represent like components.
The following description presents several examples to convey the concepts of the inventive subject matter. Although examples are used, they do not in any way limit the scope of the material. The presented embodiments should be interpreted to their broadest extent.
Brief Overview
A solution is described for resolving problems relating to unreliable IP fragmentation on a high speed network. IP fragmentation relates to an IP layer of a communication stack performing fragmentation of a datagram possibly transporting other protocols including TCP, UDP, ICMP, IGMP, or other protocols and matching the datagram to a Maximum Transfer Unit (MTU) of a physical media. Fragmentation is necessary given that IP deals with communication using virtual addresses and that IP needs to align the datagram transfer with the actual physical media which typically has a fixed maximum size. Datagrams can have a length of up to 65,535 bytes where an 802.3 Ethernet physical layer typically has a maximum size of 1500 bytes. IP performs both the fragmentation of the datagram during a transmit operation and the reassembly of the datagram upon reception. For example, the process to perform reassembly is done through an association of fields in an IPv4 header including the source IP Address (32 bits), datagram identification field (16 bits), flags (3 bits), and fragment offset field (13 bits).
Multiple datagrams in-flight are identified through a datagram identifier which has a length of 16 bits in IPv4. Typically, an IP layer in a communication stack uses a counter which increments by one every time a datagram identifier is used. The value of the counter represents the datagram identifier. The datagram corruption problem arises in reassembly; the IP layer will retain fragments of a datagram for reassembly for up to 120 seconds. Specifically, Windows® uses a 60 second window and Linux uses a 30 second window. A worst case scenario can be constructed to illustrate the issue.
Assume a fragmented datagram where one of the fragments is not received causing the receive buffer to attempt reassembly for the next 120 sec. Communication then continues using additional IP datagrams that also have multiple fragments. On a network supporting 1 Gbps, the datagram identifier will wrap in about 30 seconds. In particular a protocol using a payload size of 512 bytes would encounter a wrap in about 270 milliseconds. Considering that the IP layer waits for up to 120 sec, there is ample opportunity for in-correct association of a fragment with the original datagram; thereby, resulting in corrupted data. A payload checksum can be used to attempt to negate this situation; however, a checksum field of 16 bits and using random data, the data corruption can occur using these worst case numbers every 4.8 hours. A 10 Gbps suffers the problem ten times faster.
Clearly, this situation needs to be addressed. The situation can be addressed by adding additional optional fields to the IP headers of the datagram. The optional fields, in some embodiments, comprise extended datagram identification information which can be used by hosts to increase the pool of available datagram identifiers. If the pool has sufficient size relative to the speed of the network and to the rate at which datagrams are sent, then the increased pool of available datagram identifiers will have sufficient resolution to ensure that datagram corruption will not occur. The available datagram identifiers are controlled from being reused by their sheer number.
In an alternative embodiment for protocols where each command has a distinct response, a datagram comprising a command will also include a datagram identifier in the payload. Then the response datagram will use the identifier in its response both in the payload and in the IP layer header. A host can then use an exclusion table or list to avoid the reuse of datagram identifier during the datagram's lifetime. The exclusion table can then be scrubbed based upon entries living longer than a timeout period, which will guarantee no mismatched associations. If the exclusion list becomes full, the network is perceived as too dirty. Therefore, the solution is to create a pool of datagram identifiers that has a reduced number of available datagrams that can be controlled by one host.
Datagram Identification for Internetworking Protocols
Internetworking protocols (IP) can include standardized IPs, for example IPv4 or IPv6. Even though IPv4 and IPv6 are standardized, they have problems as previously described due to how they identify datagrams for reassembly.
The IPv4 standard states in RFC 791 that identification 110 “must be unique for that source-destination pair and protocol for the time that datagram will be active in the internet system.” Although there are numerous ways to pick a value for identification 110, no guidance is offered in the standard to ensure identification 110 is unique for the lifetime of the datagram. Typically, implementations of IPv4 use a counter that is incremented by one every time a datagram is used. Clearly, identification 110 as implemented by most IPv4 implementations do not follow the standard when the lifetime of the datagram exceeds the time it takes for a wrap condition to occur. Simply put the 16-bits of information in identification 110 represents a pool of datagram identifiers that is insufficient to maintain uniqueness for a datagram lifetime.
The preceding fields comprise the mandatory components of IPv6 header 150. Additional headers are allowed by setting appropriate values to next header 161. For example, if next header 161 has a value of six, the next header will be a TCP header; if next header 161 has a value of 17, the next header will be a UDP header; or if next header 161 has a value of 44, the next header will be a fragment header. Fragment header 180 comprises the following fields used for fragmentation:
Just as in RFC 791 for IPv4, RFC 2460 states identification 170 should have a value that “must be different than that of any other fragmented packet sent recently with the same source address and destination address.” RFC 2460 suggests using multiple counters for each pair; however, it does not suggest or motivate using multiple fields for a datagram identifier. Although identification 170 comprises a 32-bit value, it is expected that as networking speeds increase, IPv6 could experience the same data corruption problem as IPv4 does at lower speeds. For example, assuming minimal fragmented IPv6 datagrams, if the line speed is 100 Gbps, then the pool of 232 values for identification 170 could be consumed in less than 180 seconds, about three minutes, approaching the lifetime of a datagram on the network. If the line speed is 1000 Gbps, then the pool of 232 values for identification 170 could be consumed in less than 18 seconds, well below the 30 to 120 second lifetime of a datagram.
Datagrams of 512 bytes were chosen in the preceding example because block-level network storage typically operates in 512 byte blocks, although multiple blocks can be concatenated into a single datagram for reading or writing to storage media. Furthermore, the contemplated communication speeds are not unrealistic because it is possible to utilize an IP as an internal communication bus where distances are small and data rates can be high.
Datagram Identifiers
In a preferred embodiment, datagram identifiers represent a much broader concept than just identification fields (identification 110 and 170) in an IP datagram header field. Datagram identifiers comprise one or more values that together allow a datagram to be uniquely identified between a source and destination during the lifetime of the datagram. The “uniqueness” of the datagram identifier does not have to be forever. In other words, after the lifetime of the datagram has been exceeded, the datagram identifier can be reused to identify any subsequently sent datagram.
Multiple Field Datagram Identifiers
Datagram identifiers include single field identifiers or multiple field identifiers. Single field identifiers are those having a single value assigned to a datagram to identify the datagram and stored in a single field. Examples of single field identifiers include the 16-bit identification 110 in the IPv4 header or the 32-bit identification 170 in the IPv6 header. Multiple field datagram identifiers represent identifiers where two or more fields combine to form the datagram identifier. For example, identification 110 could combine with another field in the header or possibly the payload to form the datagram identifier. Alternatively, a single value for the datagram identifier could be replicated multiple times within the datagram. For example, as will be discussed in a later section a single datagram identifier value could be stored in the header of an IPv4 packet as identification 110 and replicated within the payload of the datagram itself. Other fields that could be used in a datagram identifier GUIDs, checksums, or other values that increase the “uniqueness” of the datagram identifier throughout the lifetime of the datagram.
Multiple Field Datagram Identifiers: Single Valued
Replicating datagram identifier 210 within payload 230 offers several advantages. One advantage relates to applications where it is infeasible to change the internal workings of the IP module. Placing datagram identifier 210 in payload 230 provides the application control over datagram identifier 210. An additional advantage includes resolving the datagram corruption problem without necessarily modifying the standard datagram header structure. Following sections will further clarify how placing datagram identifier 210 in the payload offers greater control over the “uniqueness” of the datagram identifier. It is contemplated the value of datagram identifier could be replicated any number of times as is desired, useful, or necessary.
Multiple Field Datagram Identifiers: Multi-valued
Preferred embodiments of multi-valued fields are located in an optional field within the header of an IP datagram. Using optional header fields minimally impacts changes to existing IP modules because the fields are optional. When a modified IP module exchanges datagrams having the optional header fields with a legacy IP module, the legacy IP module should ignore any unknown options thereby resulting in legacy behavior. When a modified IP module exchanges datagram with another modified IP module that understands the optional header fields, then a higher resolution datagram identifier resolves the datagram corruption problem.
Multiple Field Datagram Identifiers: Multi-Valued IPv4 Datagram Identifier
In a preferred embodiment, an additional datagram header option defines an extended datagram identifier as illustrated in IPv4 header 420. Header length 405 has a value greater than five to indicate there are additional 32-bit words beyond the minimum of five in the header. For example, one or more of datagram fragment identification options can be added to header 420 as indicated by options 430B and 430C. Identification 410A stores at least part of the datagram identifier. Additionally, identification 410B and 410C extended the resolution of the datagram identifier. It is contemplated that type 401B or 401C could have a value of 0×1 A (Copy=0×0, Class=0×0, Number=26 (0×1 A)) to indicate the option is an extended datagram fragment identification option. Length 407B or 407C could have a value of four to indicate that total option size is four bytes. In this embodiment identification 410B or 410C offer additional 16-bits to the datagram identifier. If more resolution is required, additional options can be added. The example embodiment illustrates a datagram identifier having 48-bits comprising identification 410A, 410B, and 410C. Although this example illustrates one embodiment for extending the pool of available datagram identifiers, one skilled in the art will recognize that other possible implementations to increase the pool of available identifiers also exist. For example, rather than repeating all the option information, the length of a single option could be increased to include bytes from subsequent 32-bit words; therefore, the length field would have a value of eight bytes where two bytes come from the header identification field and six bytes come from the datagram fragment identification option fields. Consequently, the option would comprise up to 48-bits of additional resolution creating a datagram identifier of 64 bits. By changing the option length value, the datagram identifier resolution could be increased eight bits at a time.
Multiple Field Datagram Identifiers: Multi-valued IPv6 Datagram Identifier
Preferred embodiments utilizing multi-valued datagram identifiers use the existing header identification fields as the Least Significant Bits (LSB) of the datagram identifier. By placing the LSB in the existing fields, interoperability can be maintained with legacy IP modules. It is also contemplated that new optional header fields could be defined to take over the role of the identification fields. Therefore, the datagram identifier would only be stored in the new optional fields as a single field identifier.
The previous embodiments offer a solution for the datagram corruption that involves providing a pool of datagram identifiers through increasing the number of available datagram identifiers that can be assigned to a datagram. Controlling the datagram identifiers in this way resolves the datagram corruption problem. In preferred embodiments, such an approach is beneficially applied to IP layers of network communication stacks without affecting applications. However, it might be impractical for the solution to be deployed because the sheer number of IP modules deployed in the field that have to be updated. It is contemplated that communication stacks including Windows or Linux could be updated expeditiously to some degree; however, there are millions of devices having legacy stacks that would remain including routers, gateways, printers, or other devices having legacy stacks. In which case, an application level approach could be taken by device manufactures that involve multiple field datagram identifiers having a single value. A pool of datagram identifiers is created that has a reduced number of available datagram identifiers and where the datagram identifiers are controlled during the lifetime of a datagram.
Reduced Datagram Identifier Pools
Preferably when a more application centric solution is useful, the datagram identifiers are controlled through an agreement between two hosts regarding which datagram identifiers are allowed to be used. In a preferred embodiment, a first host creates or controls a pool of datagram identifiers by making sure any datagram identifier in use will not be used again for the lifetime of a datagram to which the datagram identifier is assigned. The pool has a reduced number of available datagram identifiers because some of the identifiers might currently be in use. The first host sends a second host a suggested datagram identifier drawn from the pool. The second host uses the suggested datagram identifier in a response datagram sent back to the first host. The second host encodes the suggested datagram identifier in the header and in the payload of the response datagram. The first host receives the response datagram normally. Once the datagram is reassembled, the first host reads the suggested datagram identifier from the payload of the response datagram and places it back in the pool for use again.
Reduced Datagram Identifier Pools: Lists of Allowed and Excluded Datagram Identifiers
In-flight list 550, in a preferred embodiment, represents those identifiers that are currently being used and; therefore, represent the identifiers that are excluded from being used. In that sense, list 550 comprises an exclusion list. In a preferred embodiment, list 550 comprises zero or more identifiers as represented by identifier 555A through 555M ordered by timeout. Those identifiers that are expected to time out first are nearer the head of the list than those that are most recently deployed which are placed at the tail of the list. Similar to list 530, list 550 can be implemented through a linked list.
In one embodiment, when a datagram identifier is required, the identifier is drawn from the head of list 530 and placed at the tail of list 550 as indicated by identifier 535A. Once removed from allowed list 530, the datagram identifier is controlled from being again until the lifetime of the datagram to which it is assigned has been exceeded. In a preferred embodiment, there are two conditions under which an identifier in list 550 can be removed and placed back in list 530. The first condition is that the datagram to which the identifier is assigned has been received and correctly reassembled. Up on reassembly, the payload of the datagram is checked for the identifier and the identifier is removed from list 550 and placed at the tail of list 530 as indicated by identifier 555B. The second condition is that the datagram's lifetime has been exceeded. Under such conditions, the identifier is removed from list 555A and placed at the tail of list 530 as indicated by identifier 555A. A datagram lifetime used for controlling list 550 can be determined in many ways; preferably, it is simply larger than the reassembly time within an IP module.
Consider, for example, an embodiment using IPv4 as a transport for datagrams over an IP. Initially list 530 comprises a maximum of 65,535 datagram identifiers because the IPv4 header only supports a 16-bit identification field as previously described. As datagram identifiers are used, the list becomes shuffled; however, this does not matter because the value of the identifier does not matter as long as the identifier lasts for the lifetime of the datagram. For computer workstations having sufficient memory storing a list of 65,535 identifiers is not a problem. Through using these techniques, a host create a pool of datagram identifiers that has a reduced number of available datagram identifiers that are allowed to be assigned to a datagram; thereby, resolving the data corruption problem. The resulting pool is different than the initial pool of available datagram identifier because there a different number of datagram identifiers that are actually available and the identifiers are controlled.
Although workstations typically have sufficient memory for storing 65,535 identifiers in linked lists, other systems do not have such luxury. Embedded devices have less memory and should make use of space saving techniques for conserving memory. Some embodiments use a reduced set of identifiers that can expand if necessary. For example, rather than controlling all 65,535 identifiers, a device might only use 1024 identifiers numbered 1 through 1024. If this pool is insufficient, it can be expanded by allowing identifiers 1 through 2048 or contracted as necessary if transfer rates fall. Controlling smaller lists or recycling available identifiers in this fashion offers benefits for IPv6 implementations where is could be impractical to store all possible 232 datagram identifiers.
It is contemplated that the reduced pool could have further characteristics. For example, one should note by having a first host suggest a single datagram identifier to a second host, the resulting number of available datagram identifiers is effectively one. In addition, most devices have more than one type of traffic associated with them. Therefore, one could reduce the list of allowed datagram identifier to only those having specific bit pattern, for example limiting the list to those identifiers with the most signification bits (MSB) set. The identifiers using only the LSB could be used for normal traffic. One should note that a block of identifiers could also be suggested to the second host representing a plurality of available datagram identifiers.
To further illustrate an example embodiment, in IPv4 the identification field is 16-bits. Allowed list 530 could comprise only those identifiers that have the two most significant bits set which creates a pool of 49,152 (216−214) available identifiers, leaving 16,384 (214) identifiers that are uncontrolled and available for other traffic. Preferably, the pool has a reduced number of available datagram identifiers that is a function of powers of two, for example, 2M−2N where M is the size of the identification field in bits and N has an integer value less than 32. As N approaches M, the available number could be reduced to a small number, one or even zero.
Memory 540 represents a memory associated with a computing system. In a preferred embodiment a computer or a device operating as a client comprises memory 540 and uses the lists to access a remote device. Example remote devices include network storage devices, printers, faxes, scanners, or other devices that use IP to communicate. Embodiments of memory 540 include using RAM, flash, disk drives, or other data storage memories accessible by a computing device on a network.
Reduced Datagram Identifier Pools: Example Method of Controlling Identifiers
At step 605 host A wishes to exchange messages with host B utilizing datagrams. In this example, host A expects a response to its request. Consequently, host A draws a datagram identifier from an allowed list of datagram identifiers that represents those datagram available for use. The identifier is place in an in-flight list so that the identifier is excluded from being used again for the lifetime of the datagram to which is assigned. Meanwhile at step 650, host B waits for the message from host A.
At step 610 host A prepares a message for host B. Preferably the message comprises a request that host B take some action including responding to host A's message. Host A places the identifier within the message to host B. It is contemplated that the identifier is placed in the payload of a datagram. One should note in this example embodiment the identifier is not assigned to host A's datagram, but is rather a suggestion meant for the datagram comprising host B's response.
At step 615 host A sends the message to host B over the communications link. For example, host A calls a BSD socket send( ) call to send the message over an IP network. At step 655 host B receives the datagram from host A. Preferably the message from host A is small enough that it won't be fragmented, for example, if the total datagram size is less than 576 bytes for IPv4 the datagram should not be fragmented. Host A expects a response from host B, so at step 620 host A waits for the response. One skilled in the art of network applications will recognize this scenario as being similar to a client-server exchange.
At step 660 host B forms a response to host A's message. During this step host B reads host A's message or reviews the message's datagram to find the suggested identifier. In some embodiments, the identifier is encoded in conjunction with the message's request command so the two can be found and correlated easily. In some embodiments, the datagram identifier can also be used as a transaction ID. At step 665 host B uses the identifier as the datagram identifier for the datagram comprising its response message. In the case of IPv4, host B would place the identifier in the identification field of the IPv4 header or in the case of IPv6 host B would place the identifier in the identification field of the fragment header. In addition, host B encodes the identifier in the payload of the datagram. An astute reader will note that in this embodiment host B has an IP module adapted to allow such a datagram identifier manipulation to take place.
At step 670 host B sends the datagram comprising the response message back to host A. Host A receives the datagram at step 625. At this point the datagram could be fragmented and might arrive over a period of time. Host A attempts to reassemble the datagram comprising the response message as a function of the identifier. In this situation where host A controls the use of the identifier there is no risk for datagram corruption. Further more, host A operates normally at the IP layer without requiring modification to host A's IP module because the datagram appears as a normal datagram.
In a preferred embodiment, there are two circumstances to consider associated with the received response datagram from host B: the datagram is reassembled properly, or the datagram is not reassembled. The datagram might not be reassembled because one or more fragments might have been lost. If the datagram is reassembled properly, then host A can process the response. As host A processes the response, it reads the datagram identifier encoded in the response datagram's payload. Host A now knows that the identifier can be reused because the datagram has arrived safely and the datagram's lifetime is over; consequently at step 630 host A removes the identifier from the in-flight list and places it at the tail of the allowed list. If the datagram is not reassembled properly due to loss of fragments, the datagram will be stored in the communication stack, possibly in host A's IP module, until a timeout occurs, typically on the order of 30 to 120 seconds after which the datagram is discarded. At this point, host A still must deal with the identifier for the discarded datagram. Preferably host A has a timeout associated in-flight datagram identifiers. When the timeout has been exceeded indicating the lifetime of the datagram has expired, host A removes the identifier from the in-flight list and places the identifier at the tail of the allowed list.
The in-flight list timeout of identifiers is preferably longer than the timeout associated with reassembling datagrams, possibly as long as 300 seconds. Especially preferred embodiments couple the timeouts of the in-flight list to conditions on the communications link. Some conditions have measured values including round trip time, throughput, congestion, error rates, or other measurable conditions. Other conditions might not necessarily be measured but set according to a policy. For example, the timeout might be configured by an administrator based on the speed of the network.
The reader should note the advantages the proceeding approach has for ensuring the datagram identifiers are not reused again during the lifetime of the associated datagram. One host uses application level code to manage the identifier lists without having to change the existing communication stack on the host. This is advantageous for implementations having software applications running on established operating systems including Windows or Linux where is impracticable to make changes to the IP modules. Most application developers do not have access to communication stack internals. In addition, the approach is advantageous for vendors of embedded devices where the vendor has access to the source code of the communication stack of their devices. The vendor can modify the device's IP module to operate as described. For example companies that produce network storage products can modify their products to have the described characteristics. Furthermore, by encoding the identifier information within a message, no additional messages are required to establish the unique nature of the datagram identifier. In addition, the described approach ensures compliance to IP standards where datagram identifiers are required to be unique for the lifetime of the datagram.
Especially preferred embodiments employ the described techniques for command-response protocols including storage protocols. Storage protocols include proprietary protocols including Zetera's Z-SAN protocol or standard protocols. Standard storage protocols include iSCSI, FCIP, or iFCP. The technique can be applied to those protocols using UDP or TCP as a transport layer. For example, iSCSI implementations could be modified to exchange suggested datagram identifiers to ensure TCP datagrams do not suffer from datagram corruption while maintaining conformance to the iSCSI standard.
Providing Pools of Datagram Identifiers
At step 700 a first pool of datagram identifiers is established. In a preferred embodiment, the first pool represents the pools available in response to operating according to the standard working of IP. In IPv4, the first pool comprises the 216 identifiers available for use due to the 16-bit identification field in the IPv4 header. In IPv6, the first pool comprises the 232 identifiers available for use. In both cases typically the first pool is implemented by an incrementing counter. Counter to the standards, these pools do not provide unique identifier over the lifetime of the associated datagram when datagrams are deployed on sufficiently fast communication links. A second pool that provides available datagram identifiers having stronger uniqueness characteristics resolves the problem.
To maintain backward compatibility with legacy IP modules, a preferred embodiment comprises a threshold function to determine if a second pool of datagram identifiers is useful or not. In one embodiment, a threshold function monitors the conditions of the communications between a first host and a second host over the communication link to determine if datagram identifiers should be controlled. A preferred embodiment uses a function that relates to the lifetime of a datagram to couple more closely the lifetime to the usage of datagram identifiers.
Preferred threshold function embodiments operate as a function of a timeout period or rate at which datagrams are transferred over the communication link. Especially preferred embodiments use both the timeout period and the rate. For example, the threshold function could include calculating if a wrap condition could occur in a datagram identification field for a given data rate. If the datagram transfer rate multiplied by the datagram reassembly timeout exceeds the number of available datagram identifiers, then a wrap in the identification field could result in datagram corruption. Under such conditions, the threshold function indicates a second datagram identifier pool is beneficial. It is contemplated that the threshold function could operate as a function of other values other than a rate or timeout period including congestion. Consequently, all threshold functions are contemplated. It is also contemplated that the second pool can be toggled on or off as necessary to adapt to legacy IP modules. For example, it might be beneficial to not use the second pool when working with a legacy IP module to ensure CPU cycles are not wasted.
In addition, the values that are used in the threshold function can be measured or predetermined. A measured value can be taken in real-time or over time as the system operates. For example, the rate could be measured by calculating round trip times of messages between two hosts or the rate could be measured over the entire communication link. A predetermined value includes the network bandwidth, 1 Gbps or 10 Gbps rates for Ethernet, for example.
At step 705 a decision is made regarding using the second pool of datagram identifiers. If a second pool is not useful, then at step 790 a datagram identifier is drawn from the first pool of datagram identifiers and processing continues in the traditional manner. If a second pool is useful as indicated by the threshold function, then the second pool can be instantiated.
At step 710 the second pool of datagram identifiers is created. Providing the second pool of identifiers represents establishing a set of conditions where the datagram identifiers in the second pool are controlled in a manner that ensures they are unique for the lifetime of a datagram relative to the first pool of datagram identifiers. Therefore, the second pool is different from the first pool because the number of available datagram identifiers is different or the control over the datagram identifiers is greater. Providing the second pool includes a priori modifying an IP module before run-time or altering the behavior of the IP layer at or during run-time. The creation of the second pool further includes increasing the number of available datagram identifiers in a manner that over comes the limited supply available in the first pool causing the datagram corruption problem. Increasing the number of available datagram identifiers can be achieved as previously described through the use of additional header options comprising extended datagram identifier information. The creation of the second pool also includes decreasing the number of available datagram identifiers for a datagram. Decreasing the number includes maintaining lists of allowed datagrams identifiers or suggesting a single datagram identifier or block of identifiers. In both cases increasing the number or decreasing the number of available datagram identifiers, the datagram identifiers are controlled from taking a value that causes the datagram corruption problem within the lifetime of a datagram. More than one second pool can be created for various reasons. Multiple second pools allows for establishing a pool for each host pair, for a protocol designation (TCP, UDP, ICMP, IGMP, etc. . . . ), or even for each transport protocol port. For example, all HTTP traffic on port 80 could have a pool while FTP traffic on port 21 could have a different pool. Having more than one second pool further ensures datagram identifiers are unique as required by the IP standards.
Once the second pool of datagram identifiers is established, a datagram identifier is drawn from the second pool at step 720. Drawing the datagram identifiers includes using the next identifier in a list or receiving a suggestion of a datagram identifier to use. In the case of using a list, the identifier is taken from an allowed list of identifiers or in the case of having a large pool, the identifier is generated by incrementing a counter similar to what is done in legacy IP modules.
At step 730 the datagram identifier is controlled from being reused. In preferred embodiments the identifiers are controlled from being reused during the lifetime of the datagram by ensuring the second pool has sufficient number of available datagram identifiers that an identifier can not be used within a defined timeout period associated with the datagram lifetime. In other preferred embodiments, the datagram identifier can be controlled by keeping track of which identifiers from the second pool have been used.
At step 740 the datagram identifier is assigned to a datagram. In preferred embodiments the assignment includes placing at least a portion of the datagram identifier in an identification field of an IP header. In other embodiments, a portion of the identifier is placed in optional header fields. In yet other embodiments, a portion of the identifier is place in the payload of the datagram. Once the datagram is formed and the datagram identifier is placed within the datagram, the datagram is sent to a remote host at step 750. One skilled in the art of network applications will recognize that sending a datagram can be achieved in many ways include using ICMP, IGMP, UDP, TCP, constructing a raw datagram on an IP layer, or any other type of protocols forming datagrams.
At step 755 a check is made to see if the datagram is received or not. At any time from sending to reception of the datagram, the datagram could be fragmented. In IPv4 networks, the fragmentation could occur on the network by intermediary devices including routers. In IPv6, the source of the datagram could have fragmented the datagram. If one or more fragments of the datagram have been received, the datagram is reassembled at step 760 using the datagram identifier or other information useful to identifier datagram fragments during reassembly. Other useful information includes IP address, protocol number, port assignment, or other datagram information. At step 765 a check is made to see if the datagram has been completely reassembled.
If no fragments of the datagram have arrived yet, a decision is made with respect to the lifetime of the datagram at step 775. In some embodiments the check is made at an application layer where a host is waiting for a response message. If the datagram lifetime has not been exceeded, the system continues to wait for the datagram by returning to step 755. Waiting for the datagram proceeds until either the datagram is completely reassembled or the lifetime has been exceeded.
At step 785 a check is preformed to determine if the identifier associated with the datagram has been controlled. In preferred embodiments when the datagram's payload comprises the identifier or when an optional header field is used, the datagram identifier has been controlled. If not, then the datagram is processed normally as a legacy IP module would process it. Otherwise the datagram identifier is allowed to be reused again at step 780. In some embodiments, the identifier is moved from an in-flight or excluded list to an allowed list. In other embodiments, the second pool is sufficiently large that a datagram identifier based on a large counter will not wrap allowing the identifier to be reused after the datagram lifetime has been exceeded.
Advantages
Employing the disclosed techniques offers many advantages when resolving the datagram corruption problem. In one embodiment the solution operates at an IP layer well below an application layer. An IP layer solution ensures that applications do not have to be modified to work properly. In addition, through the use of optional IP header fields there is minimal impact with respect to older devices running legacy IP modules. A legacy IP module should operate normally by ignoring unfamiliar optional headers. Also through using multiple optional headers or creating optional headers that have extended information, the system is future proofed. When data rates increase causing the pool of available datagram identifiers to become insufficient, the datagram identifiers can be extended by tacking one more optional datagram identifier fields when useful. Placing the LSB portion of the datagram identifier in the IP header identification fields ensures backward compatibility and a fall back position to legacy behavior. The use of checksums to resolve identifier collisions is no longer necessary because the datagram corruption problem is resolved; consequently, applications similar to NFS do not have to consume CPU cycles which could result in reduce throughput.
For situations where it is impracticable to make modifications to the IP modules of a set of hosts, vendors that develop networking applications or devices can also employ the disclosed techniques. Controlling the datagram identifiers that are available or in-flight, ensures the datagram identifiers are controlled from being reused under conditions that otherwise could result in datagram corruption. A vendor can modify the IP module for their products to protect datagrams and have a corresponding software module on a system that accesses the product that controls the datagram identifiers. Example software modules for network devices include device drivers.
Coupling datagram lifetimes more closely with datagram identifiers provides for greater control over the uniqueness of the identifiers. Consequently, developers can ensure compliance with existing standards and future proof against additional datagram corruption problems.
Additional Considerations
One skilled in the art of network protocols will appreciate there are additional concepts that fall within the scope of the inventive subject. Concepts include the following items, but are not limited to the following items:
Although network storage devices are referenced as examples, one should appreciate the same techniques can be applied equally well to any network device. Network devices include any device that is accessible over communication link using datagrams. Examples of network devices include printers, digital video records, media players, faxes, scanners, copiers, VoIP phones, or other devices.
Hardware
Other aspects relate to hardware associated with the inventive subject matter. It is contemplated that one could develop hardware for storing, prototyping, manufacturing, manipulating, managing, packaging, testing, physically controlling or supporting, or for other activities associated with the physical aspects of the inventive subject matter. Therefore, the inventive subject matter includes systems, methods, or apparatus for developing, producing, manufacturing, or running the hardware. In this sense, the hardware falls within the scope of the inventive subject matter.
Software
In still another aspect, it is contemplated that one could write software that would configure, simulate, or manage various aspects of the inventive subject matter and their associated infrastructure. From that perspective the inventive subject matter includes methods of writing such software, recording the software on a machine readable form, licensing, selling, distributing, installing, or operating such software on suitable hardware. Moreover, the software per se is deemed to fall within the scope of the inventive subject matter.
Thus, specific compositions and methods of resolving a datagram corruption problem using an internetworking protocol have been disclosed. It should be apparent, however, to those skilled in the art that many more modifications besides those already described are possible without departing from the inventive concepts herein. The inventive subject matter, therefore, is not to be restricted except in the spirit of the disclosure. Moreover, in interpreting the disclosure all terms should be interpreted in the broadest possible manner consistent with the context. In particular the terms “comprises” and “comprising” should be interpreted as referring to the elements, components, or steps in a non-exclusive manner, indicating that the referenced elements, components, or steps can be present, or utilized, or combined with other elements, components, or steps that are not expressly referenced.
Adams, Mark, Witchey, Nicholas J., Hull, Richard W., Khan, Adnan, Frank, Charles William, Ludwig, Thomas Earl, Baughman, Samuel K., Sharma, Rochak
| Patent | Priority | Assignee | Title |
| 10404698, | Jan 15 2016 | F5 Networks, Inc. | Methods for adaptive organization of web application access points in webtops and devices thereof |
| 10834065, | Mar 31 2015 | F5 Networks, Inc | Methods for SSL protected NTLM re-authentication and devices thereof |
| 10904320, | Apr 26 2010 | Pure Storage, Inc. | Performance testing in a distributed storage network based on memory type |
| 8259729, | Oct 30 2002 | Citrix Systems, Inc | Wavefront detection and disambiguation of acknowledgements |
| 8270423, | Jul 29 2003 | Citrix Systems, Inc | Systems and methods of using packet boundaries for reduction in timeout prevention |
| 8411560, | Jul 29 2003 | Citrix Systems, Inc | TCP selection acknowledgements for communicating delivered and missing data packets |
| 8432800, | Jul 29 2003 | Citrix Systems, Inc | Systems and methods for stochastic-based quality of service |
| 8437284, | Jul 29 2003 | Citrix Systems, Inc | Systems and methods for additional retransmissions of dropped packets |
| 8462630, | Jul 29 2003 | Citrix Systems, Inc | Early generation of acknowledgements for flow control |
| 8553699, | Oct 30 2002 | Citrix Systems, Inc. | Wavefront detection and disambiguation of acknowledgements |
| 8824490, | Jul 29 2003 | Citrix Systems, Inc. | Automatic detection and window virtualization for flow control |
| 9008100, | Oct 30 2002 | Citrix Systems, Inc. | Wavefront detection and disambiguation of acknowledgments |
| 9047242, | Apr 26 2010 | Pure Storage, Inc | Read operation dispersed storage network frame |
| 9071543, | Jul 29 2003 | Citrix Systems, Inc. | Systems and methods for additional retransmissions of dropped packets |
| 9496991, | Jul 29 2003 | Citrix Systems, Inc. | Systems and methods of using packet boundaries for reduction in timeout prevention |
| RE47411, | Aug 16 2005 | RATEZE REMOTE MGMT. L.L.C. | Disaggregated resources and access methods |
| RE48630, | Aug 16 2005 | RATEZE REMOTE MGMT. L.L.C. | Disaggregated resources and access methods |
| RE48894, | Aug 16 2005 | RATEZE REMOTE MGMT. L.L.C. | Disaggregated resources and access methods |
| Patent | Priority | Assignee | Title |
| 4422171, | Dec 29 1980 | ALPHAREL, INCORPORATED, A CORP OF CALIFORNIA | Method and system for data communication |
| 4890227, | Jul 20 1983 | Hitachi, Ltd. | Autonomous resource management system with recorded evaluations of system performance with scheduler control including knowledge learning function |
| 5129088, | Nov 30 1987 | International Business Machines Corporation | Data processing method to create virtual disks from non-contiguous groups of logically contiguous addressable blocks of direct access storage device |
| 5193171, | Dec 11 1989 | Hitachi, Ltd. | Method of managing space of peripheral storages and apparatus for the same |
| 5444709, | Sep 30 1993 | Apple Inc | Protocol for transporting real time data |
| 5506969, | Nov 29 1993 | Sun Microsystems, Inc. | Method and apparatus for bus bandwidth management |
| 5546541, | Feb 05 1991 | International Business Machines Corporation | System for routing transaction commands to an external resource manager when the target resource is not managed by the local transaction managing computer program |
| 5590124, | May 07 1993 | Apple Inc | Link and discovery protocol for a ring interconnect architecture |
| 5590276, | Jan 08 1992 | EMC Corporation | Method for synchronizing reserved areas in a redundant storage array |
| 5634111, | Mar 16 1992 | Hitachi, Ltd. | Computer system including a device with a plurality of identifiers |
| 5742604, | Mar 28 1996 | Cisco Technology, Inc | Interswitch link mechanism for connecting high-performance network switches |
| 5758050, | Mar 12 1996 | International Business Machines Corporation | Reconfigurable data storage system |
| 5758188, | Nov 21 1995 | Maxtor Corporation | Synchronous DMA burst transfer protocol having the peripheral device toggle the strobe signal such that data is latched using both edges of the strobe signal |
| 5867686, | Nov 09 1993 | RPX Corporation | High speed real-time information storage system |
| 5884038, | May 02 1997 | RPX Corporation | Method for providing an Internet protocol address with a domain name server |
| 5889935, | May 28 1996 | EMC Corporation | Disaster control features for remote data mirroring |
| 5930786, | Oct 20 1995 | ATT& GLOBAL INFORMATION SOLUTIONS COMPANY, A CORP OF MD | Method and apparatus for providing shared data to a requesting client |
| 5937169, | Oct 29 1997 | Hewlett Packard Enterprise Development LP | Offload of TCP segmentation to a smart adapter |
| 5949977, | Oct 08 1996 | AUBETA NETWORKS CORPORATION | Method and apparatus for requesting and processing services from a plurality of nodes connected via common communication links |
| 5991891, | Dec 23 1996 | AVAGO TECHNOLOGIES GENERAL IP SINGAPORE PTE LTD | Method and apparatus for providing loop coherency |
| 6018779, | Dec 15 1997 | EMC IP HOLDING COMPANY LLC | System for encapsulating a plurality of selected commands within a single command and transmitting the single command to a remote device over a communication link therewith |
| 6081879, | Nov 04 1997 | RPX Corporation | Data processing system and virtual partitioning method for creating logical multi-level units of online storage |
| 6101559, | Oct 22 1997 | Hewlett Packard Enterprise Development LP | System for identifying the physical location of one or more peripheral devices by selecting icons on a display representing the one or more peripheral devices |
| 6105122, | Feb 06 1998 | TERADATA US, INC | I/O protocol for highly configurable multi-node processing system |
| 6128664, | Oct 20 1997 | Fujitsu Limited | Address-translating connection device |
| 6157935, | Dec 17 1996 | Qualcomm Incorporated | Remote data access and management system |
| 6157955, | Jun 15 1998 | Intel Corporation | Packet processing system including a policy engine having a classification unit |
| 6181927, | Feb 18 1997 | Nortel Networks Limited | Sponsored call and cell service |
| 6202060, | Oct 29 1996 | Qualcomm Incorporated | Data management system |
| 6246683, | May 01 1998 | Hewlett Packard Enterprise Development LP | Receive processing with network protocol bypass |
| 6253273, | Feb 06 1998 | EMC IP HOLDING COMPANY LLC | Lock mechanism |
| 6275898, | May 13 1999 | AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED | Methods and structure for RAID level migration within a logical unit |
| 6288716, | Jun 25 1997 | Samsung Electronics, Co., Ltd | Browser based command and control home network |
| 6295584, | Aug 29 1997 | SEQUENT COMPUTER SYSTEMS, INC | Multiprocessor computer system with memory map translation |
| 6330236, | Jun 11 1998 | SYNCHRODYNE NETWORKS, INC | Packet switching method with time-based routing |
| 6330615, | Sep 14 1998 | International Business Machines Corporation | Method of using address resolution protocol for constructing data frame formats for multiple partitions host network interface communications |
| 6385638, | Sep 04 1997 | HANGER SOLUTIONS, LLC | Processor resource distributor and method |
| 6396480, | Jul 17 1995 | Gateway, Inc | Context sensitive remote control groups |
| 6401183, | Apr 01 1999 | ONE WORLD LED PTY LTD | System and method for operating system independent storage management |
| 6434147, | Jan 08 1999 | Apple Inc | Method and system for sequential ordering of missing sequence numbers in SREJ frames in a telecommunication system |
| 6434683, | Nov 07 2000 | Oracle America, Inc | Method and system for transferring delta difference data to a storage device |
| 6449607, | Sep 11 1998 | Hitachi, Ltd. | Disk storage with modifiable data management function |
| 6466571, | Jan 19 1999 | UTSTARCOM, INC | Radius-based mobile internet protocol (IP) address-to-mobile identification number mapping for wireless communication |
| 6470342, | Mar 12 1999 | HEWLETT-PACKARD DEVELOPMENT COMPANY, L P | Process of maintaining a distributed map of transaction identifiers and using hashing to access these maps |
| 6473774, | Sep 28 1998 | HEWLETT-PACKARD DEVELOPMENT COMPANY, L P | Method and apparatus for record addressing in partitioned files |
| 6480934, | Sep 28 1998 | Hitachi, Ltd. | Storage control unit and method for handling data storage system using thereof |
| 6487555, | May 07 1999 | R2 SOLUTIONS LLC | Method and apparatus for finding mirrored hosts by analyzing connectivity and IP addresses |
| 6535925, | Nov 09 1999 | TELEFONAKTIEBOLAGET L M ERICSSON | Packet header compression using division remainders |
| 6549983, | May 20 1998 | SAMSUNG ELECTRONICS CO , LTD | Cache memory system and method for managing the same |
| 6567863, | Dec 07 1998 | Schneider Automation | Programmable controller coupler |
| 6587464, | Jan 08 1999 | Apple Inc | Method and system for partial reporting of missing information frames in a telecommunication system |
| 6601101, | Mar 15 2000 | Hewlett Packard Enterprise Development LP | Transparent access to network attached devices |
| 6601135, | Nov 16 2000 | GOOGLE LLC | No-integrity logical volume management method and system |
| 6618743, | Oct 09 1998 | OneWorld Internetworking, Inc.; ONEWORLD INTERNETWORKING, INC | Method and system for providing discrete user cells in a UNIX-based environment |
| 6629264, | Mar 30 2000 | Hewlett Packard Enterprise Development LP | Controller-based remote copy system with logical unit grouping |
| 6681244, | Jun 09 2000 | VALTRUS INNOVATIONS LIMITED | System and method for operating a network adapter when an associated network computing system is in a low-power state |
| 6693912, | Jun 04 1999 | Oki Electric Industry Co., Ltd. | Network interconnecting apparatus and active quality-of-service mapping method |
| 6701431, | Jan 28 2000 | Infineon Technologies AG | Method of generating a configuration for a configurable spread spectrum communication device |
| 6701432, | Apr 01 1999 | Juniper Networks, Inc | Firewall including local bus |
| 6710786, | |||
| 6711164, | Nov 05 1999 | Nokia Corporation | Method and apparatus for performing IP-ID regeneration to improve header compression efficiency |
| 6728210, | Dec 21 1998 | NEC Corporation Of America | Multi-logical access for a serial data link |
| 6732171, | May 31 2002 | Hewlett Packard Enterprise Development LP | Distributed network storage system with virtualization |
| 6732230, | Oct 20 1999 | AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED | Method of automatically migrating information from a source to an assemblage of structured data carriers and associated system and assemblage of data carriers |
| 6741554, | Aug 16 2002 | Google Technology Holdings LLC | Method and apparatus for reliably communicating information packets in a wireless communication network |
| 6754662, | Aug 01 2000 | RPX CLEARINGHOUSE LLC | Method and apparatus for fast and consistent packet classification via efficient hash-caching |
| 6757845, | Nov 30 2000 | BITMICRO LLC | Method and apparatus for testing a storage device |
| 6772161, | Dec 19 2001 | Hewlett Packard Enterprise Development LP | Object-level migration in a partition-based distributed file system |
| 6775672, | Dec 19 2001 | Hewlett Packard Enterprise Development LP | Updating references to a migrated object in a partition-based distributed file system |
| 6775673, | Dec 19 2001 | Hewlett Packard Enterprise Development LP | Logical volume-level migration in a partition-based distributed file system |
| 6795534, | Sep 04 2000 | NEC Corporation | Data recording system for IP telephone communication |
| 6799244, | Dec 13 1999 | Hitachi, Ltd. | Storage control unit with a volatile cache and a non-volatile backup cache for processing read and write requests |
| 6799255, | Jun 29 1998 | EMC IP HOLDING COMPANY LLC | Storage mapping and partitioning among multiple host processors |
| 6834326, | Feb 04 2000 | Hewlett Packard Enterprise Development LP | RAID method and device with network protocol between controller and storage devices |
| 6853382, | Oct 13 2000 | Nvidia Corporation | Controller for a memory system having multiple partitions |
| 6854021, | Oct 02 2000 | International Business Machines Corporation | Communications between partitions within a logically partitioned computer |
| 6862606, | May 11 2001 | JPMORGAN CHASE BANK, N A , AS SUCCESSOR AGENT | System and method for partitioning address space in a proxy cache server cluster |
| 6876657, | Dec 14 2000 | AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED | System and method for router packet control and ordering |
| 6882637, | Oct 14 1999 | Nokia Technologies Oy | Method and system for transmitting and receiving packets |
| 6886035, | Aug 02 1996 | Hewlett Packard Enterprise Development LP | Dynamic load balancing of a network of client and server computer |
| 6894976, | Jun 15 2000 | NetApp, Inc | Prevention and detection of IP identification wraparound errors |
| 6895461, | Apr 22 2002 | Cisco Technology, Inc | Method and apparatus for accessing remote storage using SCSI and an IP network |
| 6895511, | Oct 29 1998 | INFOBLOX INC | Method and apparatus providing for internet protocol address authentication |
| 6901497, | Oct 27 2000 | DROPBOX INC | Partition creating method and deleting method |
| 6904470, | Mar 26 2003 | EMC IP HOLDING COMPANY LLC | Device selection by a disk adapter scheduler |
| 6907473, | Oct 30 1998 | VIRNETX INC | Agile network protocol for secure communications with assured system availability |
| 6912622, | Apr 15 2002 | Microsoft Technology Licensing, LLC | Multi-level cache architecture and cache management method for peer-to-peer name resolution protocol |
| 6917616, | Sep 18 1998 | Alcatel Canada Inc | Method and apparatus for reduction and restoration of data elements pertaining to transmitted data packets in a communications network |
| 6922688, | Jan 23 1998 | PMC-SIERRA, INC | Computer system storage |
| 6928473, | Sep 26 2000 | Microsoft Technology Licensing, LLC | Measuring network jitter on application packet flows |
| 6941555, | Nov 05 1998 | Oracle International Corporation | Clustered enterprise Java™ in a secure distributed processing system |
| 6947430, | Mar 24 2000 | International Business Machines Corporation | Network adapter with embedded deep packet processing |
| 6985956, | Nov 02 2000 | Oracle America, Inc | Switching system |
| 7039934, | Dec 10 1999 | Sony Corporation | Recording system |
| 7051087, | Jun 05 2000 | Microsoft Technology Licensing, LLC | System and method for automatic detection and configuration of network parameters |
| 7065579, | Jan 22 2001 | Oracle America, Inc | System using peer discovery and peer membership protocols for accessing peer-to-peer platform resources on a network |
| 7069295, | Feb 14 2001 | The Escher Group, Ltd. | Peer-to-peer enterprise storage |
| 7073090, | Apr 23 1993 | EMC Corporation | Remote data mirroring system having a remote link adapter |
| 7111303, | Jul 16 2002 | International Business Machines Corporation | Virtual machine operating system LAN |
| 7120666, | Oct 30 2002 | RIVERBED TECHNOLOGY LLC | Transaction accelerator for client-server communication systems |
| 7145866, | Mar 01 2001 | EMC IP HOLDING COMPANY LLC | Virtual network devices |
| 7146427, | Apr 23 2002 | AVAGO TECHNOLOGIES INTERNATIONAL SALES PTE LIMITED | Polling-based mechanism for improved RPC timeout handling |
| 7149769, | Mar 26 2002 | Mosaid Technologies Incorporated | System and method for multi-destination merge in a storage area network |
| 7152069, | Oct 15 2002 | Network Appliance, Inc. | Zero copy writes through use of mbufs |
| 7184424, | Nov 12 2002 | SERVSTOR TECHNOLOGIES, LLC | Multiplexing storage element interface |
| 7188194, | Apr 22 2002 | Cisco Technology, Inc | Session-based target/LUN mapping for a storage area network and associated method |
| 7200641, | Dec 29 2000 | EMC IP HOLDING COMPANY LLC | Method and system for encoding SCSI requests for transmission using TCP/IP |
| 7203730, | Feb 13 2001 | NetApp, Inc | Method and apparatus for identifying storage devices |
| 7206805, | Sep 09 1999 | Oracle International Corporation | Asynchronous transcription object management system |
| 7237036, | Oct 14 1997 | ALACRITECH, INC | Fast-path apparatus for receiving data corresponding a TCP connection |
| 7260638, | Jul 24 2000 | Bluesocket, Inc. | Method and system for enabling seamless roaming in a wireless network |
| 7263108, | Aug 06 2002 | CAVIUM INTERNATIONAL; MARVELL ASIA PTE, LTD | Dual-mode network storage systems and methods |
| 7278142, | Aug 24 2000 | CLOUDING CORP | Dynamic computing environment using remotely allocable resources |
| 7296050, | Jan 18 2002 | MARQETA, INC | Distributed computing system and method |
| 7327735, | Nov 27 2002 | WSOU Investments, LLC | System and method for detecting lost messages transmitted between modules in a communication device |
| 7333451, | Oct 18 1999 | Apple Inc | Buffer management for mobile internet protocol |
| 7353266, | Nov 30 2000 | Microsoft Technology Licensing, LLC | System and method for managing states and user context over stateless protocols |
| 7406523, | Nov 21 2000 | Microsoft Technology Licensing, LLC | Client-server communications system and method using a semi-connectionless protocol |
| 7415018, | Sep 17 2003 | RPX Corporation | IP Time to Live (TTL) field used as a covert channel |
| 7428584, | Oct 07 2002 | Hitachi, LTD | Method for managing a network including a storage system |
| 7436789, | Oct 09 2003 | SRI International | Ad Hoc wireless node and network |
| 7447209, | Mar 09 2004 | UNIVERSITY OF NORTH CAROLINA, THE | Methods, systems, and computer program products for modeling and simulating application-level traffic characteristics in a network based on transport and network layer header information |
| 7463582, | Apr 14 2000 | U S BANK NATIONAL ASSOCIATION | System and method for scaling a two-way satellite system |
| 7526577, | Sep 19 2003 | Microsoft Technology Licensing, LLC | Multiple offload of network state objects with support for failover events |
| 7535913, | Jun 06 2002 | Nvidia Corporation | Gigabit ethernet adapter supporting the iSCSI and IPSEC protocols |
| 7742454, | Dec 16 2004 | International Business Machines Corporation | Network performance by dynamically setting a reassembly timer based on network interface |
| 20010020273, | |||
| 20010026550, | |||
| 20010049739, | |||
| 20020026558, | |||
| 20020029286, | |||
| 20020039196, | |||
| 20020052962, | |||
| 20020062387, | |||
| 20020065875, | |||
| 20020087811, | |||
| 20020091830, | |||
| 20020126658, | |||
| 20020165978, | |||
| 20030018784, | |||
| 20030023811, | |||
| 20030026246, | |||
| 20030065733, | |||
| 20030069995, | |||
| 20030081592, | |||
| 20030118053, | |||
| 20030130986, | |||
| 20030161312, | |||
| 20030172157, | |||
| 20030182349, | |||
| 20030202510, | |||
| 20030204611, | |||
| 20040025477, | |||
| 20040047367, | |||
| 20040078465, | |||
| 20040100952, | |||
| 20040181476, | |||
| 20040184455, | |||
| 20040213226, | |||
| 20050033740, | |||
| 20050058131, | |||
| 20050102522, | |||
| 20050144199, | |||
| 20050166022, | |||
| 20050175005, | |||
| 20050198371, | |||
| 20050246401, | |||
| 20050267929, | |||
| 20050270856, | |||
| 20050286517, | |||
| 20060036602, | |||
| 20060077902, | |||
| 20060133365, | |||
| 20060168345, | |||
| 20060176903, | |||
| 20070101023, | |||
| 20070110047, | |||
| 20070230476, | |||
| 20080181158, | |||
| 20080279106, | |||
| CN1359214, | |||
| EP485110, | |||
| EP654736, | |||
| EP700231, | |||
| EP706113, | |||
| JP2001094987, | |||
| JP2001359200, | |||
| JP2002252880, | |||
| JP2002318725, | |||
| JP2005265914, | |||
| WO101270, | |||
| WO215018, |
| Executed on | Assignor | Assignee | Conveyance | Frame | Reel | Doc |
| May 15 2006 | RATEZE REMOTE MGMT. L.L.C. | (assignment on the face of the patent) | / | |||
| Jun 13 2006 | ADAMS, MARK | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 14 2006 | WITCHEY, NICHOLAS J | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 14 2006 | KHAN, ADNAN | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 14 2006 | SHARMA, ROCHAK | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 14 2006 | HULL, RICHARD W | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 15 2006 | FRANK, CHARLES WILLIAM | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 19 2006 | BAUGHMAN, SAMUEL K | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 20 2006 | LUDWIG, THOMAS EARL | Zetera Corporation | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 017828 | /0538 | |
| Jun 15 2007 | Zetera Corporation | CORTRIGHT FAMILY TRUST, DATED MAY 13, 1998 | SECURITY AGREEMENT | 019453 | /0845 | |
| Jul 11 2007 | Zetera Corporation | THE FRANK REVOCABLE LIVING TRUST OF CHARLES W FRANK AND KAREN L FRANK | SECURITY AGREEMENT | 019583 | /0681 | |
| Oct 01 2007 | Zetera Corporation | WARBURG PINCUS PRIVATE EQUITY VIII, L P | SECURITY AGREEMENT | 019927 | /0793 | |
| Apr 15 2008 | Zetera Corporation | RATEZE REMOTE MGMT L L C | ASSIGNMENT OF ASSIGNORS INTEREST SEE DOCUMENT FOR DETAILS | 020866 | /0888 | |
| Apr 18 2008 | CORTRIGHT FAMILY TRUST, DATED MAY 13, 1998 | Zetera Corporation | RELEASE BY SECURED PARTY SEE DOCUMENT FOR DETAILS | 020824 | /0215 | |
| Apr 18 2008 | WARBURG PINCUS PRIVATE EQUITY VIII, L P | Zetera Corporation | RELEASE BY SECURED PARTY SEE DOCUMENT FOR DETAILS | 020824 | /0074 | |
| Apr 18 2008 | THE FRANK REVOCABLE LIVING TRUST OF CHARLES W FRANK AND KAREN L FRANK | Zetera Corporation | RELEASE BY SECURED PARTY SEE DOCUMENT FOR DETAILS | 020823 | /0949 |
| Date | Maintenance Fee Events |
| Sep 24 2014 | M1551: Payment of Maintenance Fee, 4th Year, Large Entity. |
| Sep 13 2018 | M1552: Payment of Maintenance Fee, 8th Year, Large Entity. |
| Nov 28 2022 | REM: Maintenance Fee Reminder Mailed. |
| May 15 2023 | EXP: Patent Expired for Failure to Pay Maintenance Fees. |
| Date | Maintenance Schedule |
| Apr 12 2014 | 4 years fee payment window open |
| Oct 12 2014 | 6 months grace period start (w surcharge) |
| Apr 12 2015 | patent expiry (for year 4) |
| Apr 12 2017 | 2 years to revive unintentionally abandoned end. (for year 4) |
| Apr 12 2018 | 8 years fee payment window open |
| Oct 12 2018 | 6 months grace period start (w surcharge) |
| Apr 12 2019 | patent expiry (for year 8) |
| Apr 12 2021 | 2 years to revive unintentionally abandoned end. (for year 8) |
| Apr 12 2022 | 12 years fee payment window open |
| Oct 12 2022 | 6 months grace period start (w surcharge) |
| Apr 12 2023 | patent expiry (for year 12) |
| Apr 12 2025 | 2 years to revive unintentionally abandoned end. (for year 12) |